【摘要】 利用CHARLS数据,首次引入机器学习Boosting算法——GBM(Generalized Boosted Model)的倾向评分加权法与传统的PSM(Propensity Score Model)模型,评估农地流转对家庭收入与贫困线的比值、人均收入、市民化程度和社会保障需求等农民福利指标的平均净效应。结果表明:参与流转可以显著提高农户的人均收入,帮助家庭远离贫困线,但流出土地的农户家庭收入和人均收入提高幅度更高;流出农地的农户家庭的市民化程度会显著提高,而流入农地家庭的市民化程度有所降低,但并不显著;发生流转的农户养老保险需求都会增长,二者医疗保险需求变化方向相反,租入土地农户医疗保险需求降低。更多还原

【Abstract】 This paper first introduced boosting algorithm of machine learning: Propensity Score Weighting based Generalized Boosted Model( GBM) and Propensity Score Model( PSM) to estimate the treatment effect of rural land transfer relative to baseline values by using changes in the ratio of household income to poverty line,per capital income,urbanization and social security needs.Research indicates that land circulation significantly increases the per capita income of farmers,which helps families stay away from the poverty line.Besides,the increase in the ratio of household income to poverty line and per capital income of farmers who lease out land is higher than that of farmers who rent land; The degree of citizenization of the households out of the farmland is significantly improved.However,for farmers who rent land the degree is insignificantly reduced; While the demand for social endowment insurance of both kinds of farmers increase,medical insurance needs for farmers who rent in land and those who rent out land change in the opposite direction.Farmers who rent land reduce their demand for medical insurance.更多还原